LWT - Food Science and Technology 142 (2021) 110989

Contents lists available at ScienceDirect

LWT

journal homepage: www.elsevier.com/locate/lwt

Differences in moisture sorption characteristics and browning of lesser ( diaperinus) ingredients

He Sun a, Ornella Necochea Velazco a, Catriona Lakemond a, Matthijs Dekker a, Lee Cadesky b, Maryia Mishyna a,* a Food Quality & Design Group, Wageningen University & Research, 6700 AA, Wageningen, the Netherlands b Protifarm Processing B.V., Harderwijkerweg 141B, 3852 AB, Ermelo, the Netherlands

ARTICLE INFO ABSTRACT

Keywords: Lesser mealworm (Alphitobius diaperinus) is an species that can be reared on an industrial scale for human Browning consumption. In this study, the hydration behavior of processed lesser mealworm food ingredients (whole Lesser mealworm powder, protein concentrate, textured protein) was analyzed in comparison to whey protein concentrate and Modelling tofu. Moisture adsorption isotherms were determined gravimetrically over 0.11–0.96 water activity range at 5 Proteins ◦ and 20 C, and seven sorption models were applied to fit the experimental data. The Guggenheim-Anderson-de Sorption isotherms Boer model performed best on the isotherms of all studied samples. The determined adsorption isotherms were type III according to Brunauer classification. Mealworm protein concentrate presented the strongest hydration ◦ capacity at 20 C, followed by whey protein concentrate and whole mealworm powder. Textured mealworm ◦ protein and tofu exhibited similar hydration behavior at 5 C. Moreover, browning and agglomeration of mealworm protein concentrate powder were observed at water activities of 0.73 and higher, which can be explained by its composition and porous structure and were not typical for whey protein concentrate. The moisture adsorption results are important for the prediction of lesser mealworm ingredients shelf-life or pre­ vention of undesirable quality changes during storage, but more research on microbial growth and sensory quality is needed.

1. Introduction isolates, and oils that can be used in the formulation of novel insect-based foods in order to attract consumers. This has been suc­ The world population is expected to increase to 9.7 billion in 2050 cessfully demonstrated in the lesser mealworm (Alphitobius diaperinus) of (United Nations, 2019), this comes with an increasing demand for ani­ which food ingredients products are available in Western Europe. mal protein, limited land area, and low sustainability of existing food Regarding proteins, the lesser mealworm has high protein content and production systems. In this regard, the studies for new protein sources its essential amino acid content is comparable to proteins (Yi are emerging over recent years (Fasolin et al., 2019). Edible are et al., 2013). drawing more attention and are considered as an alternative to con­ Water content and its interaction with food components usually ventional meat products due to their nutritional value and sustainability determines the stability of a food system (Rahman, 2010). Moisture of production, leading to less greenhouse gas emissions and reduced sorption reflects the relationship between moisture content and water land and water usage (van Huis, & Tomberlin, 2017). Although insects activity at given temperature and pressure. It plays an important role in have been used as traditional foods in some tropical countries (DeFo­ physicochemical and biological changes of food products (e.g., brown­ liart, 1992), their consumption is still not fully accepted by many ing, lipid oxidation, and microbial growth (Caballero-Ceron,´ western societies due to people’s natural negative attitude towards Guerrero-Beltran,´ Mújica-Paz, Torres, & Welti-Chanes, 2015)). Various them. So far, many efforts have been made to promote the edible insect types of food products often exhibit different properties. For instance, market and to increase acceptance (Shockley, Allen, & Gracer, 2017). powder products tend to absorb water and form agglomerates especially One successful strategy is processing insects into powder, protein in humid conditions, which could impair not only the functional

* Corresponding author. E-mail address: [email protected] (M. Mishyna). https://doi.org/10.1016/j.lwt.2021.110989 Received 21 July 2020; Received in revised form 24 January 2021; Accepted 25 January 2021 Available online 29 January 2021 0023-6438/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). H. Sun et al. LWT 142 (2021) 110989 properties but also lower the quality of final products (Aguilera, del insect powder (Janssen, Vincken, Van Den Broek, Fogliano, & Lake­ Valle, & Karel, 1995). Different forces and the intermolecular structure mond, 2017), a Kp of 5.60 - for insect protein concentrate and textured of food components are involved in the agglomeration phenomenon insect protein (Janssen et al., 2017), and a Kp of 6.25 - for WPC and tofu (Zafar, Vivacqua, Calvert, Ghadiri, & Cleaver, 2017). Therefore, un­ (Khatib, Aramouni, Herald, & Boyer, 2002; Rouch, Roupas, & Roginski, derstanding the characteristics of food ingredients and their moisture 2007). The content of carbohydrate was calculated by subtracting the sorption properties is a prerequisite for package selection and better contents of protein, fat, and ash from 100%. Each measurement was preservation. It is also essential for product design, process optimization, performed in triplicate. and for further understanding of the sorption mechanisms in food products (Al-Muhtaseb, McMinn, & Magee, 2002). 2.3. Particle size distribution There are numerous studies on the moisture sorption behavior of conventional food systems under different conditions. For instance, it The particle size distribution of three powder products (IP, IPC, and has been shown that increasing temperature decreased the equilibrium WPC) was determined via sieve classification with some modifications moisture content of fish meal (Vivanco & Taboada, 1998). Singh et al. (Strange & Onwulata, 2002). Test sieves with pore sizes of 1.250, 1.000, reported that type II desorption isotherms of smoked sausages 0.425, 0.315, and 0.180 mm and FRITSCH sieve shaker (Germany) were were best fitted with Halsey equation and pore size increased when used. The particle size distribution was presented as a mass fraction on a rising temperatures (Singh, Rao, Anjaneyulu, & Patil, 2001), while dry basis. Sawhney et al. also determined type II sorption isotherms for whey protein concentrate, but the isotherms were best fitted with GAB 2.4. Surface morphology ◦ equation at 25, 35, and 45 C (Sawhney, Sarkar, Patil, & Sharma, 2014). Tham et al. observed lactose crystallization in three infant formulas at The surface morphology of IP, IPC, TIP, WPC, and tofu was measured 0.428 water activity and caking in water activity range of 0.330–0.753 by Magellan 400 scanning electron microscopy (United States). Critical (Tham, Wang, Yeoh, & Zhou, 2016). However, the relevant studies for point drying method was used to dehydrate tofu, the other samples were ◦ edible insects are limited. One of few available studies on insects dried in an oven at 31 C under vacuum for 48 h. All samples were then demonstrated that cricket powder had higher hydration capacity than coated with 12 nm of tungsten and analyzed in the scanning electron black soldier fly larvae, attributing to compositional and physicochem­ microscopy with a beam current of 13 pA, 2.00 kV. The magnificationof ical differences and affecting its shelf-life (Kamau et al., 2018). Azzollini images was from 250× to 10000×. et al. found that blanching treatment reduced the hygroscopicity of yellow mealworm powder and GAB model well described its hydration 2.5. Color analysis behavior (Azzollini, Derossi, & Severini, 2016). However, to the best of our knowledge, the hydration behavior of processed lesser mealworm The color of IPC and WPC was measured by Hunterlab color flex has not been studied yet, but it is important to get insight for preventing (Virginia, United States). The total color difference (ΔE) of equilibrated the occurrence of undesirable reactions and ensuring the stability of samples at each water activity was compared with the initial ones after insect ingredients during storage. the firstdrying. Browning index (BI) definedas the purity of brown color Therefore, this study aimed to obtain moisture adsorption isotherms is frequently used as an indicator of enzymatic or nonenzymatic of lesser mealworm ingredients and to determine a suitable model browning (Buera, Lozano, & Petriella, 1986; Guerrero, Alzamora, & describing these isotherms. Isotherms of lesser mealworm powder (IP), Gerschenson, 1996) and was determined according to (Lenaerts, Van lesser mealworm protein concentrate (IPC), and whey protein concen­ Der Borght, Callens, & Van Campenhout, 2018; Palou, Lopez-Malo, ◦ trate (WPC) were studied at 20 C and isotherms of IP, textured lesser Barbosa-Canovas, Welti-Chanes, & Swanson, 1999). ◦ mealworm protein (TIP) and tofu - at 5 C. WPC and tofu were included Iris V 400 (Instrument Solutions Benelux, France) was used to take for comparison to lesser mealworm ingredients. Additionally, powder the picture and analyze the color of IPC and WPC. The method type was agglomeration and color characteristics of IPC and WPC as a function of single snapshot, the lens was a Balser 25 mm, the light mode was top and water activities were investigated. bottom light. Image size was selected as the default settings (width = 2588, hight = 1942, X axis = 0, Y axis = 0). Principal component 2. Materials and methods analysis (PCA) was used to visualize the differences in color among the samples. The threshold value of color spectrum was set at 1%. 2.1. Materials 2.6. Water activity measurements Lesser mealworm (Alphitobius diaperinus) products IP, IPC, and TIP were supplied by Protifarm (Ermelo, the Netherlands). IP represented Novasina LabMaster-aw (Lachen, Switzerland) was used to measure 100% whole powder obtained from blanched, dried, and grinded lesser the water activities of IP, IPC, TIP, WPC, and tofu before the experiment, . IPC is high protein processed powder and TIP is coagulated as well as the water activities of studied samples during moisture ◦ protein product from lesser mealworm. All products were produced in adsorption experiment. The temperature was set at 25 C. The stable accordance with HACCP guidelines. WPC was purchased from Bulk­ observation time of a chamber was 1 min (Aw) and 2 min (temperature). powders (Colchester, England), tofu - from Jumbo supermarket The measurements were performed in quadruplicate. (Wageningen, the Netherlands). The salts used were reagent grade and purchased from Sigma-Aldrich (the Netherlands). Thymol (2-isopropyl- 2.7. Moisture adsorption isotherms 5-methylphenol, ≥98.5%) was purchased from Sigma-Aldrich (the ◦ Netherlands), petroleum ether 40–60 C (PEC grade) - from Actu-All As the storage recommended temperatures vary for products with Chemicals (the Netherlands). different water activities, the adsorption isotherms were determined at ◦ ◦ 20 C for IP, IPC, WPC, and at 5 C for TIP and tofu. The isotherm of IP at ◦ 2.2. Proximate composition analysis 5 C was also determined to study the temperature effect. Isotherms were obtained by static gravimetric method (Caballero-Ceron,´ ˜ The content of dry matter, crude protein, crude fat, and ash were Guerrero-BeltrA¡n, Mújica-Paz, Torres, & Welti-Chanes, 2015). Nine measured for IP, IPC, TIP, WPC, and tofu (Fig. S1) according to AOAC saturated salt solutions (LiCl, CH3COOK, MgCl2, K2CO3, Mg(NO3)2, (Horwitz, 1975). Different nitrogen-to-protein conversion factors (Kp) NaCl, SrCl2, KCl, KNO3) with a water activity range of 0.11–0.96 were were used to calculate protein content. A Kp of 4.76 was used for whole used (Greenspan, 1977; Kiranoudis, Maroulis, Tsami, & Marinos-Kouris,

2 H. Sun et al. LWT 142 (2021) 110989

1993). The water activity of saturated salt solutions at different tem­ homoscedastic residual distribution indicates a good fit of data. peratures (Caballero-Ceron´ et al., 2015; Greenspan, 1977; Lopez-Malo,´ Palou, & Argaiz, 1994) and the specificationsof preparing saturated salt 2.9. Data analysis solutions are presented in Table S1. Salt solutions with the excess salt were stirred once a day for fivedays to assure the right formation of the The significance test (α = 0.05) was performed via LSD and Tukey saturated solution and then transferred into desiccators two days before post hoc test in IBM SPSS statistics 22. In the case of small number of the experiment. samples, Shapiro-Wilk test (α = 0.05) was applied to check the normality An amount of 1.00 ± 0.01 g sample was weighed into pre-weighed ◦ distribution of moisture adsorption data. The extended uncertainties (U) aluminium plates (W1) and dried at 55 C for 24 h that was deter­ of EMC at each water activity were calculated by the following formula mined sufficient by preliminary tests. The weight of sample and plate (Bell, 2001; Bui, Labat, & Aubert, 2017): was recorded (W ) after drying. Each desiccator contained 4 dried 2 √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ samples. Crystalline thymol was placed inside the desiccators with 2 2 U = k ∗ uA + uB (4) Aw>0.60 to inhibit microbial growth. To ensure the accuracy, one sample was taken as an indicator for each desiccator. The weight and Aw where u represents random uncertainty (Type An uncertainty), u of the indicator was measured after every 2 days until the difference in A B represents systematic uncertainty (Type An uncertainty). Assuming that weight between two consecutive measurements was about ±0.001 g the results were normally distributed, and coverage factor k = 2 to give a representing that samples were at the equilibrium state. Then the level of confidence of approximately 95%. u was calculated by the measurements for the other samples were performed. Considering the A following formula (Bell, 2001; Bui et al., 2017): atmospheric effect on sample during weighing, the operation was done as quickly as possible, generally within 30 s (Kiranoudis et al., 1993). s uA = √̅̅̅ (5) The weight of equilibrated sample and plate was recorded (W3). The n equilibrium moisture content (EMC) (g/100 g dry basis) was calculated by the following formula: where s is the standard deviation, n is the number of measurements. As the EMC value was calculated by Equation (4), so the systematic un­ W3 W2 Equilibrium ​ moisture ​ content ​ (EMC) = ∗100. (1) certainty of EMC includes the uncertainties of 3 different mass mea­ W2 W1 surements (W3, W2, and W1) and the uncertainty of dry matter content The isotherm was made by plotting EMC against corresponding measurements (d). uB was calculated by summing the partial derivatives water activities. For each isotherm, 27 experimental data points were of each parameter (Bell, 2001; Bui et al., 2017): √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ used to create the plot. √ ( √∑ Water binding capacity (WBC) related to high moisture systems √ ∂W 2 uB = ∗u(xi)) (6) (Chen, Piva, & Labuza, 1984; Wallingford & Labuza, 1983) was deter­ i ∂xi mined to compare the capacity of different types of products to bind ◦ water. WBC data were taken as the EMC values at the Aw = 0.95 (20 C) The same gravimetric scale (METTLER TOLEDO AE200, United ◦ and Aw = 0.96 (5 C), respectively (Chen et al., 1984; Wallingford & States) was used to measure W3, W2, and W1 for the whole experiments. Labuza, 1983). Applying Equations (1) and (6) to this case leads to: √[̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅(̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅) ] ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅( ) ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ 2 W W 2 W W 2 u = + 2 3 ∗u (W)2 + 2 3 ∗u (d)2 2.8. Mathematical modelling of sorption isotherms B 2 2 2 B 2 B (7) W1∗d W1∗d W1∗d

Experimental data were fitted to 7 sorption models Brunauer- uB(W) represents the uncertainty of the scale which was calculated Emmett-Teller (BET), Guggenheim-Anderson-de Boer (GAB), Kuhn, by the following formula (Bui et al., 2017): Smith, Caurie, Halsey, and Oswin that were frequently applied ( )2 ( )2 2 uDisp uLin (Al-Muhtaseb et al., 2002; Caballero-Ceron´ et al., 2015; Caurie, 1981; uB(W) = √̅̅̅ + √̅̅̅ (8) 3 3 Kamau et al., 2018; Smith, 1947). The mathematical equations of each model and the corresponding range of water activity are presented in where uDisp (display resolution) and uLin (linearity) are 0.0001 and Table S2. The model parameters were estimated by non-linear regres­ 0.0003, respectively, found in the instruction of scale. sion to minimize the residual sums of squares using SolverAid, Microsoft Excel 365 ProPlus software (Salter & De Levie, 2002). Akaike informa­ 3. Results and discussion tion criterion (AIC) was used to compare different models and evaluate the goodness of fitof each model. Due to a small number of experiments, 3.1. Proximate composition a corrected version (AICc) was used by the equation below (Burnham & Anderson, 2002): The results of proximate composition and water activities (Table 1) ) n AIC = n ∗ ln S2 + 2∗(p + 1)∗ shows that IP, IPC, and WPC were low moisture content products c (2) n p (95.5–98.7% dry matter content, Aw = 0.146–0.206), while TIP and tofu were high in moisture content (23.9–28.4% dry matter content, Aw = 2= where n is the number of data points, S SSres/n with SSres the residual 0.971–0.972). The proximate composition of IP was comparable to the sums of squares, p is the number of model parameters. The term AIC composition of lesser mealworm powder (58.4% protein, 26.3% fat, and Δ differences ( AICc) was used to interpret the quality of model. The 5.4% ash) previously described (Janssen et al., 2017). As the nutrients of Δ & equation of AICc was listed below (Burnham Anderson, 2002): larvae are affected by diets and development state (van Broekhoven, & AICc = AICci AICmin (3) Oonincx, van Huis, van Loon, 2015), there might be variations in its composition between studies. As shown in Table 1, IPC has a higher where AICci is the AICc value of each model, AICmin is the minimum content of fat (16.8%) and carbohydrate (23.8%), but lower protein value of all. The model with the minimum AICc indicates the best one. content (53.9%) in comparison to WPC (1.0% fat, 16.7% carbohydrate, Models with 0<ΔAICc<4 have substantial support to be considered as a and 78.5% protein). Similarly, higher fat and carbohydrate content, but good model (Burnham & Anderson, 2002). In addition, residuals plot lower protein content were also found in TIP compared to tofu. Although was used to evaluate the quality of the model. A random and IPC and TIP were expected to be similar to WPC and tofu in terms of

3 H. Sun et al. LWT 142 (2021) 110989

Table 1 that IPC and WPC had different surface properties. The clumping of IPC Proximate composition and water activity of IP, IPC, TIP, WPC, and tofu. powder was also observed on microphotographs, e.g., the spherical- IP IPC TIP WPC Tofu shaped IPC particles aggregated with each other (Fig. 2 and b), result­ ing in large agglomerated particles. WPC particles presented a “cell” Composition, % dry matter1 98.7 ± 96.7 ± 28.4 ± 95.5 ± 23.9 ± shape of different sizes and were relatively dispersed under SEM (Figs. 2 0.0a,2 0.1b 0.5d 0.2c 0.4e and 3b). TIP and tofu were similar in surface morphology (Figs. 2, 4a protein 46.7 ± 0.5c 53.9 ± 47.9 ± 78.5 ± 59.4 ± and 5a), both exhibited a few pores. 0.2bc 3.2c 0.2a 7.3b fat 27.9 ± 0.2a 16.8 ± 8.3 ± 0.7c 1.0 ± 5.0 ± 0.9b 0.1e 0.2d 3.4. Analysis of color ash 3.5 ± 0.0c 5.5 ± 3.0 ± 3.9 ± 5.5 ± 0.0a 0.1d 0.0b 0.0a Browning of IPC samples was observed at higher water activities 3 ± ± ± ± ± Carbohydrate 21.8 23.8 40.8 16.7 30.1 (Aw>0.73) after around 8 days, while the color of WPC samples did not 0.7bc 1.0bc 2.6a 0.1d 7.1b Aw 0.146 ± 0.206 ± 0.972 ± 0.189 ± 0.971 ± change significantly. In particular, BI values of IPC over the water ac­ 0.006d,4 0.03b 0.001a 0.001c 0.001a tivity range of 0.73–0.95 (SrCl2–KNO3) were much higher than that of IPC at lower water activities (Fig. 3), that suggests that evident 1 Dry matter content was determined on fresh basis, the other results were browning phenomenon occurred over this range. No significant differ­ presented as g/100 g dry basis. 2 Mean ± standard deviation, n = 3, the same superscript letters in the same ences in the BI values were found for WPC, indicating that no evident row are not significantly different (p > 0.05). browning of WPC has occurred. 3 The content of carbohydrate was calculated by subtracting the contents of In Fig. 4, the samples in PC1 direction show the largest variation, and protein, fat, and ash from 100%, 4Mean ± standard deviation, n = 4. PC2 axis is the second most important direction which is orthogonal to PC1 (Kassambara, 2016). The data points of IPC at Aw ranging from 0.73 product type, respectively, they slightly differed in compositions. The to 0.95 were grouped and separated from the other groups (Fig. 4a), i.e. – higher carbohydrate content of these insect products is partly explained the color of IPC samples over this water activity range (0.73 0.95) was by the fact that insects contain amounts of chitin (Janssen et al., 2017). significantly different from the others. It supports the observed brown­ ing in IPC at higher water activities (0.73–0.95). As for WPC, the overall distribution of data points was relatively close (Fig. 4b). Also, the 3.2. Particle size distribution samples after drying and with KNO3 solution treatment were found similar in PC1 direction. This means no evident color change occurred in The result of particle size distribution of three powder products (IP, WPC. It can be also seen from Fig. 4a, that the texture of IPC had IPC, and WPC) (Fig. 1) shows that IP and IPC particles were mostly changed a lot. The particles of IPC at higher water activities (0.73–0.95) distributed (76 and 70%, respectively) above 1.250 mm, which means aggregated and texture of the one at Aw of 0.95 became more liquid like, both of them were made up of relatively large particles. While IPC while the others remained powder state. However, caking of WPC was contained much smaller sized particles (less than 0.425 mm) compared observed only at Aw of 0.95 (Fig. 4b). to IP (above 0.425 mm). In contrast, half of WPC consisted of small sized Azzollini et al. also observed color alternation in freeze-dried yellow – particles (0.180 0.315 mm) with the largest one being less than 1.000 mealworm larvae (Azzollini et al., 2016), assuming that the enzymatic mm. The difference in particle size of IPC and WPC can be partly browning as a result of hydration of enzymes caused the color change of explained by clumping of IPC powder which was observed during the the sample, while non-enzymatic browning could also occur due to the movement of the sieve and shaking machine and higher fat content in absorption of moisture. Labuza found that non-enzymatic browning IPC comparing to WPC. began when monolayer moisture was formed, and reaction rates could increase to a maximum at Aw of 0.70 (Labuza, 1977). Therefore, 3.3. Surface morphology non-enzymatic browning reaction likely explains the browning in this study, considering the inactivation of enzymes caused by heat treatment ◦ IP, IPC, TIP, WPC, and tofu were observed under scanning electron temperatures higher than 80 C during the sample preparation. This idea microscopy (SEM). IP was made up of large pieces of insect tissues with a is supported by the composition of IPC with protein and also a sub­ few cracks (Figs. 2 and 1a), but no pores were observed on the outer stantial carbohydrate content. In particular, the higher carbohydrate surface (Figs. 2 and 1b). The surface of IPC had a porous microstructure content of IPC compared to WPC could accelerate the process of with a large number of pores (Fig. 2 and a), while less porous surface browning and adsorption of water. The role of Maillard reaction in structure was observed for WPC particles (Figs. 2 and 3a), suggesting browning of IPC and the presence of reducing sugars in lesser mealworm

Fig. 1. Particle size distribution (mass fraction, %) of IP, IPC, and WPC.

4 H. Sun et al. LWT 142 (2021) 110989

Fig. 2. Scanning electron microscopy (SEM) photographs of IP (1a: 1000×, 1b: 250×), IPC (2a: 2000×, 2b: 8000×), WPC (3a: 1000×, 3b: 250×), TIP (4a: 1000×, 4b: 500×), and tofu (5a: 2000×, 5b: 10000×). All samples were dehydrated and coated with tungsten before the analysis.

5 H. Sun et al. LWT 142 (2021) 110989

◦ Fig. 3. Browning index of IPC and WPC after incubation for 8 days at 0.11–0.96 water activity (Aw) range at 20 C measured with Iris V 400.

◦ Fig. 4. PCA plots of IPC (a) and WPC (b) at 0.11–0.96 water activity (Aw) range at 20 C based on Iris color analysis. Photos show IPC and WPC powders and corresponding saturated salt solutions. Lines of the same colors connect the points of a sample at a particular Aw. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) powders are of interest for future studies. Additionally, the porous 3.5. Moisture adsorption isotherms structure of IPC with a larger surface area could increase the reaction ◦ rate (Stoklosa, Lipasek, Taylor, & Mauer, 2012). The adsorption data at 5 and 20 C are listed in Table S3. The

6 H. Sun et al. LWT 142 (2021) 110989

◦ ◦ Fig. 5. Moisture adsorption isotherms at 20 C (a) and 5 C (b). isotherms of all samples obtained (Fig. 5) were type III according to Table 2 Brunauer classification (Brunauer, 2007). Isotherms of this type are ◦ ◦ Water binding capacity data at 20 C and 5 C determined by moisture relatively rare but frequently found in products containing high carbo­ adsorption isotherm. hydrate content (Blahovec & Yanniotis, 2009). The use of pre-drying method or the properties of products resulted in the lack of data Aw Water binding capacity, g water/100g solids ◦ ◦ ◦ points at water activities below 0.1, which could influence the deter­ 0.95 IP 20 C IPC 20 C WPC 20 C 25.6 ± 2.0b, 1-2 49.0 ± 5.6a 36.0 ± 6.4b mination of isotherm shape. It was also found in pineapple pulp powder ◦ ◦ ◦ 0.96 IP 5 C TIP 5 C Tofu 5 C that some moisture remained after drying, which led to a lack of data a b c 22.1 ± 0.3 17.1 ± 0.4 15.6 ± 0.4 points below Aw of 0.1 (Vigano´ et al., 2012). Therefore, optimization of 1 Mean ± standard deviation, n = 3. pre-drying method is needed to determine the isotherm shapes at low 2 Aw values. The same superscript letters in the same row are not significantly different ◦ (p > 0.05). At 20 C IPC had the highest equilibrium moisture content at the same water activity value, followed by WPC and IP (Fig. 5a). These three isotherm curves differed significantly as water activity increased. IPC found that more water was absorbed in the potato chips with a higher & exhibited the strongest water-binding capacity, followed by WPC and IP percentage of porosity (Yang, Martin, Richardson, Wu, 2017). (Table 2), that suggests that at room temperature, IPC could absorb and Therefore, porous microstructure of IPC might explain its large amount bind much more water vapor than WPC and IP. Stoklosa et al. observed of absorbed water. Also, the changes in texture and composition of IPC that smaller glass beads gained more weight than larger sized particles due to agglomeration and browning formation could affect its moisture during water adsorption (Stoklosa et al., 2012). The increased absorp­ adsorption. Although WPC contains the smallest sized particles, lack of tion of moisture with decreasing particle size was also reported for porous structure might result in less amount of absorbed water than that crystalline sucrose (Roge´ & Mathlouthi, 2000). Besides, Yang et al. of IPC. As for IP, large particle size and lack of porous structure caused

7 H. Sun et al. LWT 142 (2021) 110989 the least absorption of moisture content. 3.8. Uncertainties ◦ At 5 C, the isotherms of IP, TIP, and tofu were similar at lower water activities as shown in Fig. 5b. But IP differed from TIP and tofu in The uncertainties (U, uA, uB) of IP at each water activity condition adsorption behavior as water activity increased. At higher water activ­ are listed in Table S5. It has been assumed that all experimental data ities (Aw>0.60), the equilibrium moisture content of IP was much were normally distributed, although some data were not according to higher than that of TIP and tofu at the same Aw value. This could be Shapiro-Wilk test, that could be explained by the small sample size. The linked to the different interactions between water molecules and food uncertainties were relatively low, except for the ones at higher water ◦ components in IP, TIP, and tofu (Caballero-Ceron´ et al., 2015). Besides, activities. For instance, the U values of IPC and WPC at 20 C were about IP differed form TIP and tofu in structure, which could also lead to the 6.48 and 7.27, respectively, at the highest water activity (Aw = 0.95, difference in moisture sorption behavior. No significant difference was KNO3). It was also observed that the uA value generally determined the found in the isotherms of TIP and tofu, which suggests that these two extended uncertainty U, as the systematic uncertainty uB was much products share similar moisture adsorption properties and could be lower (Table S5). This result was comparable to the evaluation of un­ explained by their similar surface morphology. certainties on moisture content of (Bui et al., 2017).

3.9. Applications 3.6. Effect of temperature on adsorption isotherms

◦ The moisture sorption isotherms are frequently used to predict the The adsorption isotherms of IP fittedwith GAB model at 5 and 20 C stability of food systems. Labuza created a “stability map” describing are shown in Fig. 5. At the constant water activity, the equilibrium various physiochemical and biological reactions that could occur in food moisture content (EMC) increased with a higher temperature. This systems as a function of water activity (Labuza, 1977). Information means more water molecules were absorbed and bound by IP when related to these reactions, such as the rate and occurrence water activity increasing temperature. However, the observed result was contrary to range could be gained from this map. The observation in our study that the general theory that increasing temperature would decrease the browning of IPC occurs at Aw>0.73 that is in agreement with the fact, binding energy between molecules and reduce the number of active sites that the non-enzymatic browning could occur above 0.20 water activity for water adsorption, thus leading to an increase of water activity due to (Labuza, 1977). Thus, it is possible to correlate and predict undesirable activated water molecules (Caballero-Ceron´ et al., 2015). Kamau et al. reactions in lesser mealworm ingredients based on the moisture observed a significant decrease of adsorption for cricket and black adsorption isotherms. Besides, Tonon et al. combined the glass transition soldier flylarvae powder with increasing temperature, which was more temperatures with water sorption isotherms and determined the critical evident in black soldier fly larvae (Kamau et al., 2018). Vivanco and water activity and glass transition temperature for food storage (Tonon Taboada also found a decrease in moisture content of fish meal with et al., 2009). Therefore, the isotherms could also be used for preventing increasing temperature (Vivanco & Taboada, 1998). This general tem­ undesirable structural or texture changes in food products. Labuza and perature effect on moisture sorption was also found in many other food Altunakar also proposed a model to predict the shelf life of food products products, for instance, sprayed dried tomato pulp (Goula, Karapantsios, (Labuza & Altunakar, 2007). The model calculated parameters can be Achilias, & Adamopoulos, 2008), sesame flour (Menkov & Durakova, used to estimate the products shelf life by considering initial and equi­ 2007), and macadamia nuts (Palipane & Driscoll, 1993). However, librium moisture content, critical moisture content together with Saravacos and Sinchfield observed an increase in the amount of water packaging material parameters (permeability, thickness, water vapor absorbed by freeze-dried starch-glucose gel and peach as the tempera­ ◦ transmission rate). In general, packaging of dry powdery and tofu-like ture rose from 20 to 30 C (Saravacos & Stinchfield, 1965). Similar insect-derived food ingredients should meet the requirements for behavior was also found on sorption isotherms of raisins (Saravacos, limiting the water vapor, oxygen, and light transmission. Based on the Tsiourvas, & Tsami, 1986), suggesting that the increase of sugar disso­ results presented concrete shelf life extending measures with regard to lution at higher temperature became the determining factor, which agglomeration and browning can be taken by measures that keep aw reduced the opposite effect of temperature on the sorption of non-sugar below 0.73. solids (Saravacos et al., 1986). 4. Conclusions 3.7. Modelling IP, IPC, TIP, WPC, and tofu exhibited type III adsorption moisture Seven sorption models were fitted to experimental data. The esti­ isotherms according to Brunauer classification in this work. The GAB mated parameters of models and results of Akaike information criterion model best described the data in all isotherms in comparison to other are listed in Table S4. The minimum AICc value indicates the best quality models based on its minimum value of the Akaike information criterion. of fitting.As shown in the table, the GAB model obtained the lowest AICc The random residual distributions of GAB model also proved that the value of all seven models. The adsorption isotherms fitted with GAB experimental data fitted well to GAB model. IPC absorbed the largest ◦ model and corresponding residual plots were calculated. GAB model quantity of water at 20 C probably due to its smaller particle size and well described the isotherms of all 5 samples at corresponding temper­ porous structure, followed by WPC and IP. TIP and tofu exhibited similar ◦ atures, although it overestimated the adsorption data at lower water moisture adsorption properties at 5 C. Evident browning and texture ◦ ◦ activities (Aw<0.50), for instance IP at 20 and 5 C, IPC at 20 C (Fig. 5). changes were only found in IPC at higher water activities, that could be Further evaluation of the goodness of fit was done by analyzing the explained by high carbohydrate content and porous structure. To gain residual plots and the parameter correlation matrices and uncertainties, more information about storage conditions or to predict shelf life, more this analysis also indicated that the GAB model is appropriate for the research should be performed and combined with moisture sorption data. Considering all factors, the GAB model well described the isotherm results, for instance the studies on microbial growth, physical adsorption isotherms and provided the best fit of all. The Caurie and changes or consumer sensory test during storage. Oswin models described experimental data reasonably well (figures were not shown), however the ΔAICc values of Caurie and Oswin were CRediT authorship contribution statement both higher than 10, indicating that these two models failed to explain some substantial explainable variation in the data (Burnham & Ander­ He Sun: Investigation, Formal analysis, Writing - original draft, son, 2002). Thus, Caurie and Oswin were excluded from further Writing - review & editing, Visualization. Ornella Necochea Velazco: consideration in this case. Conceptualization, Writing - review & editing, Supervision. Catriona

8 H. Sun et al. LWT 142 (2021) 110989

Lakemond: Methodology, Writing - review & editing. Matthijs Dekker: Horwitz, W. (1975). Official methods of analysis. Washington, DC: Association of Official Methodology, Validation, Writing - review & editing. Lee Cadesky: Analytical Chemists. van Huis, A., & Tomberlin, J. K. (Eds.). (2017). and feed: From production to Conceptualization, Writing - review & editing. Maryia Mishyna: consumption. The Netherlands: Wageningen Academic Publishers. https://doi.org/ Conceptualization, Writing - review & editing, Supervision. 10.3920/978-90-8686-849-0. Janssen, R. H., Vincken, J. P., Van Den Broek, L. A. M., Fogliano, V., & Lakemond, C. M. M. (2017). Nitrogen-to-protein conversion factors for three edible Declaration of competing interest insects: Tenebrio molitor, Alphitobius diaperinus and Hermetia illucens. Journal of Agricultural and Food Chemistry, 65, 2275–2278. https://doi.org/10.1021/acs. There is no competing conflict among all authors. jafc.7b00471 Kamau, E., Mutungi, C., Kinyuru, J., Imathiu, S., Tanga, C., Affognon, H., et al. (2018). Moisture adsorption properties and shelf-life estimation of dried and pulverised Acknowledgements edible house cricket Acheta domesticus (L.) and black soldier fly larvae Hermetia illucens (L.). Food Research International, 106, 420–427. https://doi.org/10.1016/J. FOODRES.2018.01.012 The author would like to thank Wageningen Electron Microscopy Kassambara, A. (2016). Practical guide to principal component methods in R. Centre for the scanning electron microscopy analysis. Khatib, K. A., Aramouni, F. M., Herald, T. J., & Boyer, J. E. (2002). Physicochemical characteristics of soft tofu formulated from selected soybean varieties. Journal of Food Quality, 25, 289–303. https://doi.org/10.1111/j.1745-4557.2002.tb01026.x Appendix A. Supplementary data Kiranoudis, C. T., Maroulis, Z. B., Tsami, E., & Marinos-Kouris, D. (1993). Equilibrium moisture content and heat of desorption of some vegetables. Journal of Food Supplementary data to this article can be found online at https://doi. Engineering, 20, 55–74. https://doi.org/10.1016/0260-8774(93)90019-G Labuza, T. P. (1977). The properties of water in relationship to water binding in foods: A org/10.1016/j.lwt.2021.110989. review. Journal of Food Processing and Preservation, 1, 167–190. Labuza, T. P., & Altunakar, L. (2007). Water activity prediction and moisture sorption References isotherms. In G. V. Barbosa-Canovas,´ A. J. Fontana Jr, S. J. Schmidt, & T. P. Labuza (Eds.), Water activity in foods: Fundamentals and applications (pp. 109–154). Oxford, UK: Blackwell Publishing Ltd. https://doi.org/10.1002/9780470376454.ch5. Aguilera, J. M., del Valle, J. M., & Karel, M. (1995). Caking phenomena in amorphous Lenaerts, S., Van Der Borght, M., Callens, A., & Van Campenhout, L. (2018). Suitability of food powders. Trends in Food Science & Technology, 6(5), 149–155. https://doi.org/ microwave drying for mealworms (Tenebrio molitor) as alternative to freeze drying: 10.1016/S0924-2244(00)89023-8 Impact on nutritional quality and colour. Food Chemistry, 254, 129–136. https://doi. Al-Muhtaseb, A. H., McMinn, W. A. M., & Magee, T. R. A. (2002). Moisture sorption org/10.1016/j.foodchem.2018.02.006 isotherm characteristics of food products: A review. Food and Bioproducts Processing, Lopez-Malo,´ A., Palou, E., & Argaiz, A. (1994). Measurement of water activity of 80(2), 118–128. https://doi.org/10.1205/09603080252938753 saturated salt solutions at various temperatures. In International symposium on the Azzollini, D., Derossi, A., & Severini, C. (2016). Understanding the drying kinetic and properties of water (pp. 113–116). hygroscopic behaviour of larvae of yellow mealworm (Tenebrio molitor) and the Menkov, N. D., & Durakova, A. G. (2007). Moisture sorption isotherms of sesame flourat effects on their quality. Journal of Insects as Food and Feed, 2(4), 233–243. https:// several temperatures. Food Technology and Biotechnology, 45, 96–100. doi.org/10.3920/JIFF2016.0001 Palipane, K. B., & Driscoll, R. H. (1993). Moisture sorption characteristics of in-shell Bell, S. (2001). A beginneraˆ€™s guide to uncertainty of measurement. Teddington, macadamia nuts. Journal of Food Engineering, 18(1), 63–76. https://doi.org/ Middlesex, United Kingdom. 10.1016/0260-8774(93)90075-U Blahovec, J., & Yanniotis, S. (2009). Modified classification of sorption isotherms. Palou, E., Lopez-Malo, A., Barbosa-Canovas, G. V., Welti-Chanes, J., & Swanson, B. G. Journal of Food Engineering, 91(1), 72–77. https://doi.org/10.1016/j. (1999). Polyphenoloxidase activity and color of blanched and high hydrostatic jfoodeng.2008.08.007 pressure treated banana puree. Journal of Food Science, 64(1), 42–45. https://doi. van Broekhoven, S., Oonincx, D. G. A. B., van Huis, A., & van Loon, J. J. A. (2015). org/10.1111/j.1365-2621.1999.tb09857.x Growth performance and feed conversion efficiency of three edible mealworm Rahman, M. S. (2010). Food stability determination by macro-micro region concept in species (Coleoptera: Tenebrionidae) on diets composed of organic by-products. the state diagram and by defininga critical temperature. Journal of Food Engineering, Journal of Insect Physiology, 73, 1–10. https://doi.org/10.1016/j. 99(4), 402–416. https://doi.org/10.1016/j.jfoodeng.2009.07.011 jinsphys.2014.12.005 Roge,´ B., & Mathlouthi, M. (2000). Caking of sucrose crystals: Effect of water content and Brunauer, S. (2007). The adsorption of gases and vapors vol I - physical adsorption (Vol. 1). crystal size. Zuckerindustrie, 125(5), 336–340. Brunauer Press. Rouch, D. A., Roupas, P., & Roginski, H. (2007). True protein value of milk and dairy Buera, M., Lozano, R., & Petriella, C. (1986). Definition of color in the non-enzymatic products. Australian Journal of Dairy Technology, 62(1), 26–30. browning process. Farbe, 32, 318–322. Salter, C., & De Levie, R. (2002). Nonlinear fits of standard curves: A simple route to Bui, R., Labat, M., & Aubert, J.-E. (2017). Comparison of the saturated salt solution and uncertainties in unknowns. Journal of Chemical Education, 79(2), 268. https://doi. the dynamic vapor sorption techniques based on the measured sorption isotherm of org/10.1021/ed079p268 barley straw. Construction and Building Materials, 141, 140–151. https://doi.org/ Saravacos, G., & Stinchfield, R. (1965). Effect of temperature and pressure on the 10.1016/j.conbuildmat.2017.03.005 sorption of water vapor by freeze-dried food materials. Journal of Food Science, 30, Burnham, K. P., & Anderson, D. R. (2002). Model selection and multimodel inference. A 779–786. practical information-theoretic approach. New York: Springer-Verlag. ˜ Saravacos, G., Tsiourvas, D., & Tsami, E. (1986). Effect of temperature on the water Caballero-Ceron,´ C., Guerrero-BeltrA¡n, J. A., Mújica-Paz, H., Torres, J. A., & Welti- adsorption isotherms of sultana raisins. Journal of Food Science, 51, 381–383. Chanes, J. (2015). Moisture sorption isotherms of foods: Experimental methodology, Sawhney, I. K., Sarkar, B. C., Patil, G. R., & Sharma, H. K. (2014). Moisture sorption mathematical analysis, and practical applications. In Water stress in biological, isotherms and thermodynamic properties of whey protein concentrate powder from chemical, pharmaceutical and food systems (pp. 187–214). New York: Springer. buffalo skim milk. Journal of Food Processing and Preservation, 38(4), 1787–1798. https://doi.org/10.1007/978-1-4939-2578-0_15. https://doi.org/10.1111/jfpp.12148 Caurie, M. (1981). Derivation of full range moisture sorption isotherms. In Shockley, M., Allen, R. N., & Gracer, D. (2017). Product development and promotion. In L. B. Rockland, & G. F. Stewart (Eds.), Water activity: Influences on food quality. A Insects as food and feed: From production to consumption (pp. 398–427). Wageningen treatise on the influence of bound and free water on the quality and stability of foods and Academic Publisher. other natural products (pp. 63–87). New York: Academic Press. Singh, R. R. B., Rao, K. H., Anjaneyulu, A. S. R., & Patil, G. R. (2001). Moisture sorption Chen, J. Y., Piva, M., & Labuza, T. P. (1984). Evaluation of water binding capacity (WBC) properties of smoked chicken sausages from spent hen meat. Food Research of food fiber sources. Journal of Food Science, 49, 59–63. https://doi.org/10.1111/ International, 34(2-3), 143–148. https://doi.org/10.1016/S0963-9969(00)00145-9 j.1365-2621.1984.tb13668.x Smith, S. E. (1947). The sorption of water vapor by high polymers. Journal of the DeFoliart, G. R. (1992). Insects as human food. Gene DeFoliart discusses some nutritional American Chemical Society, 69(3), 646–651. https://doi.org/10.1021/ja01195a053 and economic aspects. Crop Protection, 11(5), 395–399. https://doi.org/10.1016/ Stoklosa, A. M., Lipasek, R. A., Taylor, L. S., & Mauer, L. J. (2012). Effects of storage 0261-2194(92)90020-6 conditions, formulation, and particle size on moisture sorption and flowability of Fasolin, L. H., Pereira, R. N., Pinheiro, A. C., Martins, J. T., Andrade, C. C. P., powders: A study of deliquescent ingredient blends. Food Research International, 49 Ramos, O. L., et al. (2019). Emergent food proteins - towards sustainability, health (2), 783–791. https://doi.org/10.1016/j.foodres.2012.09.034 and innovation. Food Research International, 125, 108586. https://doi.org/10.1016/ Strange, E. D., & Onwulata, C. I. (2002). Effect of particle size on the water sorption j.foodres.2019.108586. Elsevier Ltd. properties of cereal fibers. Journal of Food Quality, 25(1), 63–73. https://doi.org/ Goula, A. M., Karapantsios, T. D., Achilias, D. S., & Adamopoulos, K. G. (2008). Water 10.1111/j.1745-4557.2002.tb01008.x sorption isotherms and glass transition temperature of spray dried tomato pulp. Tham, T. W. Y., Wang, C., Yeoh, A. T. H., & Zhou, W. (2016). Moisture sorption isotherm Journal of Food Engineering, 85(1), 73–83. https://doi.org/10.1016/j. and caking properties of infant formulas. Journal of Food Engineering, 175, 117–126. jfoodeng.2007.07.015 https://doi.org/10.1016/j.jfoodeng.2015.12.014 Greenspan, L. (1977). Humidity fixed points of binary saturated aqueous solutions. Tonon, R. V., Baroni, A. F., Brabet, C., Gibert, O., Pallet, D., & Hubinger, M. D. (2009). Journal of Research of the National Bureau of Standards, 81(1), 89–96. ˜ Water sorption and glass transition temperature of spray dried aA§ai (Euterpe Guerrero, S., Alzamora, S. M., & Gerschenson, L. N. (1996). Optimization of a combined ˜ oleracea Mart.) juice. Journal of Food Engineering, 94(3-4), 215–221. https://doi.org/ factors technology for preserving banana purA©e to minimize colour changes using 10.1016/j.jfoodeng.2009.03.009 the response surface methodology. Journal of Food Engineering, 28(3-4), 307–322. https://doi.org/10.1016/0260-8774(95)00036-4

9 H. Sun et al. LWT 142 (2021) 110989

United Nations. (2019). World population prospects. Retrieved from https://population.un Journal of Food Science, 48(1), 1–5. https://doi.org/10.1111/j.1365-2621.1983. .org/wpp/ Accessed April 28, 2020. tb14775.x ˜ Vigano,´ J., Azuara, E., Telis, V. R. N., Beristain, C. I., JimA©nez, M., & Telis-Romero, J. Yang, J., Martin, A., Richardson, S., & Wu, C.-H. (2017). Microstructure investigation (2012). Role of enthalpy and entropy in moisture sorption behavior of pineapple and its effects on moisture sorption in fried potato chips. Journal of Food Engineering, pulp powder produced by different drying methods. Thermochimica Acta, 528, 214, 117–128. https://doi.org/10.1016/j.jfoodeng.2017.06.034 63–71. https://doi.org/10.1016/j.tca.2011.11.011 Yi, L., Lakemond, C. M. M., Sagis, L. M. C., Eisner-Schadler, V., van Huis, A., & van Vivanco, M. M., & Taboada, O. M. (1998). Thermodynamic behavior of fishmeal during Boekel, M. A. J. S. (2013). Extraction and characterisation of protein fractions from adsorption. Drying Technology, 16(9-10), 1827–1842. https://doi.org/10.1080/ five insect species. Food Chemistry, 141, 3341–3348. https://doi.org/10.1016/j. 07373939808917498 foodchem.2013.05.115 Wallingford, L., & Labuza, T. P. (1983). Evaluation of the water binding properties of Zafar, U., Vivacqua, V., Calvert, G., Ghadiri, M., & Cleaver, J. A. S. (2017). A review of food hydrocolloids by physical/chemical methods and in a low fat meat emulsion. bulk powder caking. Powder Technology, 313, 389–401. https://doi.org/10.1016/j. powtec.2017.02.024

10